Minimalistic multichip DIY LED build

Anoter thing to consider is the light output per Watt. Sure your getting more light per chip but are you becoming less effecient.

Cree XM-L-U2 at 4.8 Watts = 590 Lumns or 122 Lums per Watt
Cree XP-G-R5 at 4.9 Watts = 493 Lums or per Watt 100 Lums per Watt
Ebay 10 Watt Pro Light LED at 10 Watts = 750 lums or 75 lums per watt
Ebay 20 Watt squareness LED at 20 Watts = 1600 lumns or 80 lumns per watt
Ebay 50 Watt LED at 50 Watts = 3,000 Lumns 0r 60 Lumns per watt

Without the heat factor if we compare 20 Watt to the XML's on a 300 Watt system using 15 of the 20 Watt LED's we have 24,000 Lumns. But if we build the system with XM-L LEDs to get 24,000 Lumns we need 40 LED's each drawing 4.8 Watts for a total of 192 Watts. Your actualy saving 1/3 of the power going with multi LED's.

Now dependent on you cooling the 50 Watt LED's can use a lot more power from multi fans and or liquid cooling.
 
Lasseff,

Did I read correctly that you would prefer to use 20K 100watters at this point? You did say that they were less blue than the 10K and 16K didn't you? I am thinking about pulling the trigger on a 100w 20K from AC-RC, an Akasa Venom Voodoo CPU Cooler, and I need to nail down a good dimmable driver for it. Reading through all this and trying to figure out which driver to get is giving me a headache! :-P
 
Lasseff,

Did I read correctly that you would prefer to use 20K 100watters at this point? You did say that they were less blue than the 10K and 16K didn't you? I am thinking about pulling the trigger on a 100w 20K from AC-RC, an Akasa Venom Voodoo CPU Cooler, and I need to nail down a good dimmable driver for it. Reading through all this and trying to figure out which driver to get is giving me a headache! :-P


There are some things to consider here. First off with true Color Temperature reading 5,000K to 7,200K are generaly considered neutral whites. They should not have a noticable color tint to them compared to natural sunlight. Yet we know that when put in with Blue LED's or T-5's that a 6,500K looks yellow to most people. Simularly when we put a warm white bulb next to a 6,500K bulb it looks blue. this is the natural corection that occurs between our eyes and brain.

If you look at a Color Chart you will see that light regardless of the color temperature is not ever pure white. There is always a tint to it that usualy goes unnoticed. I had 10,000K lights on my aquarium and to mee they looked very blue when I was using incadescent room lights. But when I switched to 6,500K room lights my aquarium lost that blueish tint. Wen I had 3,200K lighting in my closet I could not tell if a pair of pants were black, or dark blue, but when I switched to 7,200K bulbs in the closet the difference became very obvious.

The next thing is that color temperature if accuratly measured will be noticable when compared to another bulb and be very predictable. The higher the color temperature the larger the ratio is between blue and red. Simularly the lower the color temperature the more red there is and the less blue there is. Green has only a little effect on actual color temperature but is very inportant if you look at CRI rating. CRI numbers are what you should see compared to being outside on a sunny day but being in the shade. A CRI of 100 would basicly be an even balance between red green and blue.

Bow the stunbler to all this how light manufacturers rate there bulbs. No two manufacturers use the exact same methods. Besides that there is a tollerance that each manufacture uses to clasify there bulbs. If Johny uses a 10% tolerance and calls a bulb 10,000 Kyou could buy two bulbs from him and one would be at 11,000K and another would be at 9,100K, if they are using true K rating for there tolerance. But if there using spectrum peaks then the diffeence could be as much as 7,500K to 12,000K. But the hard thing is when you moving up to the 20,000K plus range it only takes a slight difference in the spectrum peaks to double the K rating. In reality any rating above 16,000K is really at the discretion of the manufacturer.

If someone can find Sanjay's old test results on quality Metal Hide bulbs you will see that his tests showed that color temperature rating by manufacturers were no more than a rough guide. Some bulbs were measured to be within 200K of the advertised rating others were off by as much as 6,000K. Yet thell came within the manufacturers specifications dependent on the method used.
 
I would agree...Traditional logic would have t that a 20K bulbs would be bluer than a 16K or 10K hence the question to Lassef. I'm just not a fan of blue tanks...The nicest lit tank I ever had was with ap air of 20K 400w radiums on PFO HQI ballasts with a pair of NO actinic tubes. Nice ice white that brought out all the colors, not just blues and greens. That is what I am trying to achieve but with a 12x6 footprint to light, I'm going to be dropping a lot of money of these and I want to get as much input from those in the know before getting a test light...

Are you familiar with the best dimmable driver to run that 100w led on by any chance?
 
Lasseff,

Did I read correctly that you would prefer to use 20K 100watters at this point? You did say that they were less blue than the 10K and 16K didn't you? I am thinking about pulling the trigger on a 100w 20K from AC-RC, an Akasa Venom Voodoo CPU Cooler, and I need to nail down a good dimmable driver for it. Reading through all this and trying to figure out which driver to get is giving me a headache! :-P

I have not compare the 100 watts but I have compare 10 watts of the brand Epiled. I have 10 000 K, 16 000 K and 20 000 K.

I agree of course with that 20 000K should contain relatively more blue wavelengths than 16 000 K or 10 000 K. It is part of the definition of the Kelvin scale of light temperature.

However, when I with my eye compares the three, I find the light from that which is stated to be 20 000 K as more yellow than the other two. I have experienced this with two different batches. This is my experiences of Epiled's LED.

Epistars LED (or other manufacturer's LED) I have not compared like that, I have no idea if it is the same.

Sincerely Lasse
 
Anoter thing to consider is the light output per Watt. Sure your getting more light per chip but are you becoming less effecient.

Cree XM-L-U2 at 4.8 Watts = 590 Lumns or 122 Lums per Watt
Cree XP-G-R5 at 4.9 Watts = 493 Lums or per Watt 100 Lums per Watt
Ebay 10 Watt Pro Light LED at 10 Watts = 750 lums or 75 lums per watt
Ebay 20 Watt squareness LED at 20 Watts = 1600 lumns or 80 lumns per watt
Ebay 50 Watt LED at 50 Watts = 3,000 Lumns 0r 60 Lumns per watt

Without the heat factor if we compare 20 Watt to the XML's on a 300 Watt system using 15 of the 20 Watt LED's we have 24,000 Lumns. But if we build the system with XM-L LEDs to get 24,000 Lumns we need 40 LED's each drawing 4.8 Watts for a total of 192 Watts. Your actualy saving 1/3 of the power going with multi LED's.

Now dependent on you cooling the 50 Watt LED's can use a lot more power from multi fans and or liquid cooling.

There is no doubt that the Cree XL-M is an excellent LED and that it gives a good output However, it was not lumens per watt, I was interested when I started experimenting with multi-chips. The reason I chose the 10-watt chip was that Cree's LED was not possible to obtain in 10-20000 K. I was interested in obtaining a light source was white but that gave a lot of the effect in the blue region as I think the right wavelengths for photosynthesis is more important for corals than a concept (lumen) based on the human eye's ability to see colors and light. Then when Maglofster began experimenting with 50 and 100 watt chip, I became interested in them since I had a light source with great effect in a small space So small so that you can use a lens system for a light beam capable of penetrating into several meters of water, or have a high PAR reading over a larger area in a normal tank.

I also belong to those who believe that trying to imitate the sunlight when you have corals in most cases is not the correct way to go. Most of the corals we have in the aquarium, live in nature deeper than 4 - 5 meters and there are already many wavelengths are filtered out, especially the reds.

Sincerely Lasse
 
Last edited:
Lasse:

Have you looked at the Cree Multi Chip LED's. These can go up to 48 Watts and do have that single point output which is very bright. These are basicly an array of XML chips into one die. Not a lot different than the Chineese chips in the basic idea. But you do get a lot more light per watt.

I will agree that the Cree Does not make a so called 20,000K LED. But in reality does anyone? If you study the K scale it is bascly a balance of Red and Blue that determines the K rating. While I will agree that corals need much more Blue light than they do Red light I will also say that you can combine Blue and "white" chips to get basicly any K rating you desire.

The big think many people forget is that when you start getting up in to K scale around 20,000K these values are often theorized rather than real. If you look at Sanjays old MH specs you will see that even some of the best MH bulbs were advertised at K values up to 6,000K away from what they realy produced.

I'm with you in the believe that you only need a very minimal amount of Red Light. But if you look at the various studies on Light for Corals you will see that different corals require different frequencies. Red is one that some shallow corals need but that can also be very detrimental to many deep water corals. Unfortunatly in our Aquariums we try to keep corals from not only 2 meter but also 50 meters.

The two things I think you will be fighting with super high powered LED's is the shadowing effects and also cooling. Cooling is one reason I like Crees as they do not produce near the heat as some other that I tried.

KENTRON11

As far as the reds under 20,000K goes there are two types of red we see in the aquariums.

Florescent Reds which are actualy created by the corals when it receives one wave lenght of light and converts it to another wave lenght. The florescent reds will probably look great under 20,000K lights especialy if they are suplemented with more Blues.

Then there is the reflected Reds, which do need red light for us to see. These can be noted in fish like clowns which with a lot of red light will look bright red but with little or no red can turn marron to actualy black. If there is no RED light the object has no red light to reflect back at you to appear red. Some corals also have a lot of reflective red in them like birdsnest. I actualy added red so the my birdsnest did not look black with white tips in my pre dawn to post dusk time period.
 
Now that's something that I was throwing around...I know that in some of the Japanese show tanks spot lighting was used to accents the coloration of some corals. I wonder if that's something that could be done here. I think that reds and oranges are some of the hardest colors to really showcase with LEDs so i was thinking of a few spotlight led's to accent coral colonies when the tank starts filling in...
 
OK so how about some talk on spread...It would seem that with these one could feasibly go back to the MH mentality when it comes to positioning and the number of chips needed to light a tank. My tank is going to be appx 12 feet long, 6 feet front to back and about 5 feet deep. With those dimensions in mind, I am considering 100w LED's most likely driven at closer to 150-200w a piece with 60 degree optics and spaced out as shown in the drawing below. I'm thinking of starting with 4 chips and then filling in as needed with 50 w chips to balance out and shading.

I plan on angling the chips to focus the beam angle, keep as much light off the glass/sides as possible to curb algae growth on those surfaces. With a tank that size only viewable from the front, lighting can be applied a bit differently to accommodate viewing from the front only. The small red circle represents an area I would need a 50w chip in to cover....

The thicker gray line down the center virtually cuts the tank in half, showing the light spread from the side view with the chip mounted near the from of the tank angled backwards...Thoughts?

picture.php
 
Hi Kentrob11

I think you and I are after similar lights. I will be mostly lighting with MH but at 52" deep i was looking at getting some 100w LED cannons, until I found this thread!

I want 100w LEDs to drive up to 200-250w to pinpoint areas on the sand bed.

I have located a suitable LED and 300w heatsink I could just do with some info on the best driver to get, a dimmable one would be great!
 
Nice! It'll be neat to compare notes...I would hope that you like the LED's enough to replace your MH with them eventually :-) Which LED and heatsink are you planning on going with?
 
Lasse:

I will agree that the Cree Does not make a so called 20,000K LED. But in reality does anyone? If you study the K scale it is bascly a balance of Red and Blue that determines the K rating. While I will agree that corals need much more Blue light than they do Red light I will also say that you can combine Blue and "white" chips to get basicly any K rating you desire.

The big think many people forget is that when you start getting up in to K scale around 20,000K these values are often theorized rather than real. If you look at Sanjays old MH specs you will see that even some of the best MH bulbs were advertised at K values up to 6,000K away from what they realy produced.

Here I´m with you, its the reason why I wrote i the first that 10 - 16 000K is best according to my experiences


I'm with you in the believe that you only need a very minimal amount of Red Light. But if you look at the various studies on Light for Corals you will see that different corals require different frequencies. Red is one that some shallow corals need but that can also be very detrimental to many deep water corals. Unfortunatly in our Aquariums we try to keep corals from not only 2 meter but also 50 meters.

Must people using Cree use a ratio of 2 RB to 1 white of different K. Because that I prefeer 10 - 16 000 K I work with a ratio of 1/3 RB to 2/3 white of different K. Hence I get the red wavelengths I need and I get a lot of blue wavelengths without getting to blue light in the aquarium

The two things I think you will be fighting with super high powered LED's is the shadowing effects and also cooling. Cooling is one reason I like Crees as they do not produce near the heat as some other that I tried.

I has no proplems with shadowing because I work both with lenses and without. I do either se any problems with cooling, I have not the experiences that the LED I use is dispatching more heat than my Cree XL-M at the same wattage.

I have seen Cree multichip but has not tested them. I´m sure that we in the future will see more multichip based on Cree LED.

Sincerely Lasse
 
This LED HERE

This Heat sink rated at 300w HERE

I will use 60 degree optics to "hopefully" add a shaft of light down to the sandbed with a decent PAR. If it does work out I will build another that will use a wider optic to see what par I get vs the 400w MH.
 
I will use 60 degree optics to "hopefully" add a shaft of light down to the sandbed with a decent PAR. If it does work out I will build another that will use a wider optic to see what par I get vs the 400w MH.

Sounds like my plan as well...When are you planning on doing this?
 
Back
Top